About
Voice typing that understands the way people actually talk.
Built by a small team in Israel that needed dictation to understand Hebrew without keeping the recordings.
Why Lytly exists
Lytly started as a personal tool. We needed voice typing that understood Hebrew, handled a sentence that switches language halfway, and did not keep our conversations on someone's server. After months of building and testing, we are sharing it.
A small independent team in Israel builds it. Every dictation ends with a receipt that shows when the audio was discarded.
How it works under the hood
Lytly transcribes with a large speech model served on specialized fast-inference hardware. Audio is sent only during active recording and deleted immediately. A second provider serves as an automatic fallback if the primary provider is unavailable. Provider API keys live on our backend, never in the app.
Get in touch
Try the app, or write to us at [email protected].
What we stand for
Privacy by design
Your audio is processed in memory and discarded before the HTTP response returns. We do not want your words, only to transcribe them.
Speed without compromise
Sub-2-second round-trips, direct paste into any focused app, and silence detection that stops when you stop, not three seconds later.
Truly multilingual
Eight languages are first-class: English, French, Spanish, German, Hebrew, Russian, Chinese and Japanese. Same hotkey, same UI, same paste flow, without a "switch language" menu.